Mailing List Archive

On Friday, April 05, 1996 6:19 AM, Wolfgang Henke[SMTP:wolfgang@whnet.com] wrote:
@ salo@msc.edu (Tim Salo) wrote:
@ It would appear that there may also be a strong argument that the
@ tremendous proliferation of [small] ISPs is a significant contributor
@ to the growth of the size of the Internet routing tables.
@
@
@Tim,
@
@this smells like an opportunity to me. To my regret I've been observing
@the decline of the supercomputer industry for years due to a lack of demand
@(KSR, CM5). Can't the supercomputer gurus apply some of their extensive
@knowledge gained in a super-router?
@
@
@Wolfgang
@

Solutions that move the overall architecture toward the need for "bigger, faster, etc."
will drive the Internet down the same path as the supercomputer industry...

In my opinion, we need to sit back, look at the big picture, and come up with
incremental architectural changes. We need to RE-apply the Internet Philosophy
to the Internet itself and investigate the result.

--
Jim Fleming
UNETY Systems, Inc.
Naperville, IL 60563

e-mail: JimFleming@unety.net

>
> Solutions that move the overall architecture toward the need for "bigger, faster, etc."
> will drive the Internet down the same path as the supercomputer industry...
>
> In my opinion, we need to sit back, look at the big picture, and come up with
> incremental architectural changes. We need to RE-apply the Internet Philosophy
> to the Internet itself and investigate the result.
>
> --
> Jim Fleming
> UNETY Systems, Inc.
> Naperville, IL 60563
>
> e-mail: JimFleming@unety.net

bears repeating again......rony



Ronald Barron Yokubaitis A Texas NetWorking, Inc.
Texas Networking Inc. B TOTAL INTERNET SOLUTIONS
San Antonio 210-272 8111 5 $9.95 Internet Access
Austin 512-472 2532 L e-mail info@texas.net
J http://www.texas.net
_.. . ._ _... ..... ._.. ._ _ _ ._ _... ..... ._.. ._ _ _ _._

> Solutions that move the overall architecture toward the need for "bigger,
> faster, etc." will drive the Internet down the same path as the supercomputer
> industry...
>
> In my opinion, we need to sit back, look at the big picture, and come up with
> incremental architectural changes. We need to RE-apply the Internet Philosophy
> to the Internet itself and investigate the result.


I prefer incremental changes as well. The only problem is that they
might not be sufficient to solve the current bottlenecks. It might
work if we restrict new users just to V.34 modem speeds. But if we
have *potential* new user growth with ATT Worldnet with up to 80 million,
@home with up to 50 million and the fast growing traditional ISPs,
(which show no signs of consolidation as far as I can tell) as well as
widespread access speed increases to ISDN BRI, ADSL and cable, it may
not be enough.

As Robert Moskovitz pointed out, even the growth in common used backbones
speeds is not keeping up:

1. 56 kbps
2. 1.544 Mbps increase by 24
3. 44.736 Mbps increase by 28
4. 155.520 Mbps increase by merely 3

Just keeping in step with past growth patterns would require a step
to OC-24c at 1244.15 Mbps now, but there are no routers which come
even close to those speeds.


Wolfgang


PS. www.whnet.com/wolfgang/giga.html has an overview of routers.

On Saturday, April 06, 1996 4:20 AM, Wolfgang Henke[SMTP:wolfgang@whnet.com] wrote:
@
@> Solutions that move the overall architecture toward the need for "bigger,
@> faster, etc." will drive the Internet down the same path as the supercomputer
@> industry...
@>
@> In my opinion, we need to sit back, look at the big picture, and come up with
@> incremental architectural changes. We need to RE-apply the Internet Philosophy
@> to the Internet itself and investigate the result.
@
@
@I prefer incremental changes as well. The only problem is that they
@might not be sufficient to solve the current bottlenecks. It might
@work if we restrict new users just to V.34 modem speeds. But if we
@have *potential* new user growth with ATT Worldnet with up to 80 million,
@@home with up to 50 million and the fast growing traditional ISPs,
@(which show no signs of consolidation as far as I can tell) as well as
@widespread access speed increases to ISDN BRI, ADSL and cable, it may
@not be enough.
@
@As Robert Moskovitz pointed out, even the growth in common used backbones
@speeds is not keeping up:
@
@1. 56 kbps
@2. 1.544 Mbps increase by 24
@3. 44.736 Mbps increase by 28
@4. 155.520 Mbps increase by merely 3
@
@Just keeping in step with past growth patterns would require a step
@to OC-24c at 1244.15 Mbps now, but there are no routers which come
@even close to those speeds.
@
@
@Wolfgang
@
@
@PS. www.whnet.com/wolfgang/giga.html has an overview of routers.
@
@

Raw bandwidth and router capacity are only two dimensions of the problem.

You have not mentioned that the current "architectures" may inject a certain
amount of "overhead" which is being carried on the same network with subscriber
traffic. The "service perception" of the user has not necessarily been taken into
account in developing these architectures. This is partly because there was not
a large enough subscriber base to obtain valid usage patterns and statistics.

A simple example can be easily demonstrated when a consumer is shown that
a broken DNS system can be circumvented by using an IP address in the URL
for their web browser. This is like telling a telephone user, "if (411) Directorty
Assistance is out to lunch, just dial the number directly". Their eyes light up when
they see that the "site" is not busy, the DNS system is choking them. It is
unfortunate that the consumer has not yet been educated that they should test
the site directly, before declaring that the entire system is broken.

Now that various usage patterns are developing, it is time to look at the entire
architecture and characterize what each component is trying to do. This should
not be restricted to Web Browsing. Internet voice and audio is an important area.
Raw file transfers are also important. The maintenance infrastructure also has
to be addressed.

In my case, I am primarily interested in distributed, object-oriented, architecture-
neutral software development, distribution and billing. I am interested in seeing the
network support a "sea of objects". These smart pieces of data and information
need a clean addressing scheme to be able to flow around the world. The Legacy
Internet is a reasonably good vehicle to provide raw bit transportation for these
objects, but very little else. (except maybe some Java demo programs to keep
everyone happy until 1/1/98)

Because of the growing problems of Cats and Dogs fighting, us Dolphins need
to make sure that we do not come ashore. We just need to transport "objects"
via sonar from one coast to the other...even the U.S. Navy has not figured out
what Dolphins are saying...:-)

--
Jim Fleming
UNETY Systems, Inc.
Naperville, IL 60563

e-mail: JimFleming@unety.net

On Sat, 6 Apr 1996, Wolfgang Henke wrote:

> As Robert Moskovitz pointed out, even the growth in common used backbones
> speeds is not keeping up:
>
> 1. 56 kbps
> 2. 1.544 Mbps increase by 24
> 3. 44.736 Mbps increase by 28
> 4. 155.520 Mbps increase by merely 3
>
> Just keeping in step with past growth patterns would require a step
> to OC-24c at 1244.15 Mbps now, but there are no routers which come
> even close to those speeds.

Even backbones are not backbones anymore. Sprint, MCI et al. operate
meshes with multiple internal paths. If you have an average of 8 alternate
paths of OC3c, then you get closer to a 24x multiple of DS3. Of course
it's not really that simple, but I don't think that things are as bad
as they look in your table above.


Michael Dillon Voice: +1-604-546-8022
Memra Software Inc. Fax: +1-604-546-3049
http://www.memra.com E-mail: michael@memra.com

On Saturday, April 06, 1996 6:04 AM, Michael Dillon[SMTP:michael@memra.com] wrote:
@On Sat, 6 Apr 1996, Wolfgang Henke wrote:
@
@> As Robert Moskovitz pointed out, even the growth in common used backbones
@> speeds is not keeping up:
@>
@> 1. 56 kbps
@> 2. 1.544 Mbps increase by 24
@> 3. 44.736 Mbps increase by 28
@> 4. 155.520 Mbps increase by merely 3
@>
@> Just keeping in step with past growth patterns would require a step
@> to OC-24c at 1244.15 Mbps now, but there are no routers which come
@> even close to those speeds.
@
@Even backbones are not backbones anymore. Sprint, MCI et al. operate
@meshes with multiple internal paths. If you have an average of 8 alternate
@paths of OC3c, then you get closer to a 24x multiple of DS3. Of course
@it's not really that simple, but I don't think that things are as bad
@as they look in your table above.
@
@
@Michael Dillon Voice: +1-604-546-8022

But don't forget those "bit buckets" that are sitting there ready to handle the
social engineering needs of the net...;-)

...I wonder who has to empty those bit buckets...???...that must be a messy job...


--
Jim Fleming
UNETY Systems, Inc.
Naperville, IL 60563

e-mail: JimFleming@unety.net

you know,

i wish that i knew what any of you guys were talking about.
it gives me no end of amusement to listen to people postulate
about how we run our network. SNR is getting pretty high
in this discussion. it's beginning to sound more like a gossip
session than a discussion group.

Jeff Young
young@mci.net

> Return-Path: JimFleming@unety.net
> Return-Path: nanog-owner@merit.edu
> Received: from merit.edu (merit.edu [35.1.1.42]) by postoffice.reston.mci.net (8.6.12/8.6.6) with ESMTP id PAA05058; Sat, 6 Apr 1996 15:46:55 -0500
> Received: (from daemon@localhost) by merit.edu (8.7.5/merit-2.0) id PAA03220 for nanog-outgoing; Sat, 6 Apr 1996 15:32:15 -0500 (EST)
> Received: from doorstep.unety.net (root@usi-00-10.Naperville.unety.net [204.70.107.30]) by merit.edu (8.7.5/merit-2.0) with SMTP id PAA03215 for <nanog@merit.edu>; Sat, 6 Apr 1996 15:32:12 -0500 (EST)
> Received: from webster.unety.net (webster.unety.net [206.31.202.8]) by doorstep.unety.net (8.6.9/8.6.9) with SMTP id OAA01348; Sat, 6 Apr 1996 14:26:55 -0600
> Received: by webster.unety.net with Microsoft Mail
> id <01BB23C5.76052A00@webster.unety.net>; Sat, 6 Apr 1996 14:29:27 -0600
> Message-ID: <01BB23C5.76052A00@webster.unety.net>
> From: Jim Fleming <JimFleming@unety.net>
> To: "'Michael Dillon'" <michael@memra.com>,
> "nanog@merit.edu"
> <nanog@merit.edu>
> Subject: RE: CIDR, Sprint and the Big guys.
> Date: Sat, 6 Apr 1996 14:29:26 -0600
> MIME-Version: 1.0
> Content-Type: text/plain; charset="us-ascii"
> Content-Transfer-Encoding: 7bit
> Sender: owner-nanog@merit.edu
> Precedence: bulk
> Content-Length: 1254
>
> On Saturday, April 06, 1996 6:04 AM, Michael Dillon[SMTP:michael@memra.com] wrote:
> @On Sat, 6 Apr 1996, Wolfgang Henke wrote:
> @
> @> As Robert Moskovitz pointed out, even the growth in common used backbones
> @> speeds is not keeping up:
> @>
> @> 1. 56 kbps
> @> 2. 1.544 Mbps increase by 24
> @> 3. 44.736 Mbps increase by 28
> @> 4. 155.520 Mbps increase by merely 3
> @>
> @> Just keeping in step with past growth patterns would require a step
> @> to OC-24c at 1244.15 Mbps now, but there are no routers which come
> @> even close to those speeds.
> @
> @Even backbones are not backbones anymore. Sprint, MCI et al. operate
> @meshes with multiple internal paths. If you have an average of 8 alternate
> @paths of OC3c, then you get closer to a 24x multiple of DS3. Of course
> @it's not really that simple, but I don't think that things are as bad
> @as they look in your table above.
> @
> @
> @Michael Dillon Voice: +1-604-546-8022
>
> But don't forget those "bit buckets" that are sitting there ready to handle the
> social engineering needs of the net...;-)
>
> ...I wonder who has to empty those bit buckets...???...that must be a messy job...
>
>
> --
> Jim Fleming
> UNETY Systems, Inc.
> Naperville, IL 60563
>
> e-mail: JimFleming@unety.net
>

On Saturday, April 06, 1996 5:55 PM, Jeff Young[SMTP:young@mci.net] wrote:
@you know,
@
@i wish that i knew what any of you guys were talking about.
@it gives me no end of amusement to listen to people postulate
@about how we run our network. SNR is getting pretty high
@in this discussion. it's beginning to sound more like a gossip
@session than a discussion group.
@
@Jeff Young
@young@mci.net
@

In my opinion, the world is rapidly moving to a maturity where levels of
service and performance will be tracked, evaluated, analyzed, etc.
This will be part of the agreement(s) between the various participants.

I believe that most people up and down the Internet "food chain", just
want solid, reliable, secure service....for low cost...:-)

Because of the nature of the Internet, we all live in "Flatland" (a great
book by the way). There is no concept of "up and down" and hierarchies
are difficult to form. Instead, flat triangles are formed. It is sometimes
hard to see this world because there is no third dimension to rise above
the scene to survey the status.

In some cases, a triangle is set up where you have a "customer" on
one corner, the InterNIC on the other and an ISP on the third. The
customer could be a small ISP and the ISP could be a major carrier
but the point is the same...triangles can be deadly, nothing gets done.

...a triangle, as opposed to a hierarchy, can cause some interesting
dead-locks where the buck is always on the other corner(s) of the three-way
arrangement...

...with a hierarchy, the buck passes up and down and it is easier to
see where it is at any point in time...

In order to break the dead-locks, someone has to try to take a leadership
role in these flat triangular relationship(s). This can be difficult because the other
two parties can always be resisting the change. There may be no cooperation.

Some parties in a two-dimensional flat triangular relationship may actually
discover that they are living in the one-dimensional world of lines. In that world,
they can only see two other parties (one at a time) and there is no concept that
the other parties see each other. In that world there is no chance for cooperation.

We can only judge our perception of the state of the system via our planar
relationships with other parties that think they have a global view. It will be
interesting to see if a three-dimensional world ever gets started in the Internet.
In order for that to happen, someone has to prove that they see more than
two other parties, cooperate with all of them, and all of them cooperate in
return. That is going to be tricky...:-)

--
Jim Fleming
UNETY Systems, Inc.
Naperville, IL 60563

e-mail: JimFleming@unety.net